Understanding Latency: RPC vs. gRPC vs. Shredstreame

Not all connections are created equal. The method of data retrieval dictates the viability of any low-latency strategy.

• Standard JSON-RPC (HTTP): ~400-1000+ ms

  • Mechanism: Request-response over TCP/TLS. High overhead per request.

  • Data: Finalized or Confirmed blocks only.

  • Use Case: General dApp interaction. Unsuitable for MEV.

• Geyser (gRPC): ~50-200 ms (Public Nodes) / < 50-200 ms (Co-located)

  • Mechanism: Persistent, bidirectional streaming over gRPC. Low overhead.

  • Data: Processed, fully-formed blocks (post-facto).

  • Use Case: High-performance analytics, notifications, back-running.

• Jito Shredstream (gRPC): ~30-50 ms (via Public Relayers) / < 10-30 ms (Co-located)

  • Mechanism: Persistent, bidirectional streaming of raw block fragments.

  • Data: Pre-finalization shreds, as they are created by the leader.

  • Use Case: The absolute bleeding edge. Required for competitive front-running.

Why This Matters: The difference between Geyser and Shredstream is the difference between reacting to the past and acting on the present. For strategies aiming to front-run, access to pre-finalization data is the only way to gain the necessary temporal edge.